perf: Vectorized nested loop join for in-memory engine

[ritchie46]

Cross-join + filter and ie_join predicates that aren't equi join or ie-join can be filtered during the cross join operation, thereby often drastically reducing maximum memory usage.

supersedes #20329

[codecov]

Codecov Report

Attention: Patch coverage is 88.25503% with 35 lines in your changes missing coverage. Please review.

Project coverage is 79.04%. Comparing base (b430f64) to head (b4c3ee9).
Report is 6 commits behind head on main.

Files with missing lines	Patch %	Lines
crates/polars-ops/src/frame/join/args.rs	35.00%	13 Missing ⚠️
...rates/polars-python/src/lazyframe/visitor/nodes.rs	0.00%	11 Missing ⚠️
crates/polars-ops/src/frame/join/cross_join.rs	95.29%	4 Missing ⚠️
crates/polars-mem-engine/src/executors/join.rs	75.00%	1 Missing ⚠️
crates/polars-mem-engine/src/executors/sort.rs	0.00%	1 Missing ⚠️
crates/polars-ops/src/frame/join/mod.rs	97.56%	1 Missing ⚠️
...s/polars-pipe/src/executors/operators/reproject.rs	0.00%	1 Missing ⚠️
crates/polars-plan/src/dsl/options.rs	94.11%	1 Missing ⚠️
.../polars-plan/src/plans/optimizer/collapse_joins.rs	91.66%	1 Missing ⚠️
...lan/src/plans/optimizer/predicate_pushdown/join.rs	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #20495      +/-   ##
==========================================
+ Coverage   79.02%   79.04%   +0.01%     
==========================================
  Files        1563     1563              
  Lines      220594   220779     +185     
  Branches     2502     2502              
==========================================
+ Hits       174327   174510     +183     
- Misses      45693    45695       +2     
  Partials      574      574              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.